Mapping Intrinsic Functional Brain Changes and Repetitive Transcranial

Home , Parahippocampal gyrus, Precentral gyrus

Sleep Medicine 16 (2015) 785–791

Contents lists available at ScienceDirect

Sleep Medicine

journal homepage: www.elsevier.com/locate/sleep

Original Article Mapping intrinsic functional brain changes and repetitive transcranial magnetic stimulation neuromodulation in idiopathic restless legs syndrome: a resting-state functional magnetic resonance imaging study Chunyan Liu a,b, Zhengjia Dai c,d, Ruihua Zhang e, Mo Zhang f, Yue Hou a, Zhigang Qi f, Zhaoyang Huang a, Yicong Lin a, Shuqin Zhan a, Yong He c,d,*, Yuping Wang a,b,** a Department of Neurology, Xuan Wu Hospital, Capital Medical University, Beijing, China b Beijing Key Laboratory of Neuromodulation, Beijing, China c State Key Laboratory of Cognitive Neuroscience and Learning & IDG/McGovern Institute for Brain Research, Beijing Normal University, Beijing, China d Center for Collaboration and Innovation in Brain and Learning Sciences, Beijing Normal University, Beijing, China e Department of Functional Neurology, Lu He Teaching Hospital, Capital Medical University, Beijing, China f Department of Radiology, Xuan Wu Hospital, Capital Medical University, Beijing, China

ARTICLE INFO ABSTRACT

Article history: Objective: The objectives of this study were, first, to explore differences in brain activity between normal Received 1 October 2014 people and idiopathic restless legs syndrome (RLS) patients during asymptomatic periods; and, second, Received in revised form 11 December to determine whether administering repetitive transcranial magnetic stimulation (rTMS) to specific cor- 2014 tical regions would reverse any observed differences in brain activity and alleviate patient symptoms. Accepted 29 December 2014 Methods: Fifteen idiopathic RLS patients (nine drug-naive patients) and 14 gender- and age-matched healthy Available online 20 March 2015 controls were enrolled. Resting-state functional magnetic resonance imaging was used to measure the amplitude of low-frequency fluctuations (ALFF) in spontaneous brain activity during asymptomatic periods. Keywords: RLS Seven patients received high-frequency (5 Hz) rTMS directed toward the leg area of the primary motor ALFF cortex. Scores on the International Restless Legs Syndrome Study Group (IRLSSG) Rating Scale and ALFF rTMS values were measured before and after treatment. Resting-state functional MRI Results: Compared with healthy controls, RLS patients showed lower ALFF in the sensorimotor and visual Pathophysiology processing regions, and higher ALFF in the insula, parahippocampal and hippocampal gyri, left posteri- Neuromodulation or parietal areas, and brainstem. These results were largely conserved when only drug-naive patients were considered. After rTMS treatment, ALFF in several sensorimotor and visual regions were significantly elevated and IRLSSG Rating Scale scores decreased, indicating improved RLS symptoms. Conclusions: High-frequency rTMS delivered to the leg area of the primary motor cortex may raise functional activity in the sensorimotor and occipital regions, leading to improve symptoms in RLS patients. These results provide novel insight into RLS pathophysiology and suggest a potential mechanism for rTMS therapy in idiopathic RLS patients. © 2014 Elsevier B.V. All rights reserved.

1. Introduction which is associated with iron deﬁciency, uremia, and peripheral neuropathy. Key features include an unpleasant sensation in the Restless legs syndrome (RLS) is a sensorimotor disorder that con- lower limbs that appears or worsens during the night and disap- sists of idiopathic RLS (without known cause) and secondary RLS, pears or improves with movement [1]. Although the pathophysiology of idiopathic RLS remains incompletely understood, several studies suggest that it is related to central nervous system abnormalities

Chunyan Liu and Zhengjia Dai contributed equally to this work. [2–5]. Three self-evoked, event-related functional magnetic reso- * Corresponding author. State Key Laboratory of Cognitive Neuroscience and nance imaging (fMRI) studies reported activation in the cerebellum, Learning & IDG/McGovern Institute for Brain Research, Beijing Normal University, thalamus, brainstem, precentral gyrus, and primary somatosen- Beijing 100875, China. Tel.: +10 5880 2036; fax: +10 5880 2036. sory cortex during symptomatic periods [3–5]. The question arises E-mail address: [email protected] (Y. He). as to whether or not patterns of functional activity change during ** Corresponding author. Department of Neurology, Xuan Wu Hospital, Capital Medical University, Beijing 100053, China. Tel.: +10 8315 7841; fax: +10 8315 7841. asymptomatic periods. Resting-state fMRI, a promising neuroimaging E-mail address: [email protected] (Y. Wang). technique that noninvasively measures spontaneous/intrinsic brain http://dx.doi.org/10.1016/j.sleep.2014.12.029 1389-9457/© 2014 Elsevier B.V. All rights reserved. 786 C. Liu et al./Sleep Medicine 16 (2015) 785–791 activity [6], has been widely used to study healthy and diseased brain of the cortical regions with altered ALFF values, and we assessed function. Zang et al. [7] proposed using the amplitude of low- the effect on RLS symptoms. frequency fluctuations (ALFF; calculated as the square root of the power spectrum in a frequency range, usually 0.01–0.08 Hz) to assess the amplitude of resting-state spontaneous brain activity. By mea- 2. Methods suring the ALFF, researchers have found altered baseline brain activity in patients with attention-deficit/hyperactivity disorder [7] and post- 2.1. Subjects traumatic stress disorder [8]. These recent studies indicate that the ALFF is physiologically meaningful and reflects intrinsic or spon- Fifteen right-handed idiopathic RLS patients (12 females and three taneous neuronal activity in the brain. Thus, measuring the ALFF males; age range: 35–72 years; mean age: 56.53 ± 9.75 years; nine during asymptomatic periods and comparing it to that of control drug-naive patients) (Table 1) participated in the study. RLS was di- subjects might reveal regions of altered functional activity in RLS agnosed through clinical interview by a neurologist with sleep that may be the basis for the development of symptoms. medicine expertise (Y.H.) and according to the International Rest- Due to the augmentation of RLS symptoms during long-term less Legs Syndrome Study Group (IRLSSG) criteria [1]. Severity was treatment with dopaminergic medications, we urgently need new scored on the IRLSSG Rating Scale and the Johns Hopkins Restless therapeutic methods. Repetitive transcranial magnetic stimula- Legs Severity Scale. We also used the Pittsburgh Sleep Quality Index tion (rTMS) is a newly developed noninvasive technique that can (PSQI) to assess sleep quality. We scored patients on the Hamilton modulate brain function by improving cortical plasticity and can Anxiety Rating Scale (HAM-A) and Hamilton Depression Rating Scale be used to treat some neurological disorders [9–11]. Currently, the (HAM-D, respectively) to exclude patients with serious anxiety consensus is that high-frequency rTMS is excitatory, whereas low- (score > 21) or depression (score > 20). In addition, we excluded pa- frequency rTMS is inhibitory [12]. Here, we investigated whether tients with a history of alcohol or drug abuse, anemia, renal disease, or not excitatory rTMS could be used to therapeutically alter brain spinal cord or nerve root injury, or other neuropathies or sleep dis- function in RLS patients. First, we used resting-state fMRI to search orders. All patients had normal results on general medical and for brain regions for which the ALFF values differed between RLS neurological examinations. Routine laboratory test results (includ- patients and normal controls. Then we used rTMS to modulate one ing serum levels of hemoglobin, iron/ferritin, urea, creatinine, vitamin

Table 1 Demographic and clinical data in idiopathic RLS patients and healthy controls.

Group Sex Age (y) Disease Family Medication Agree rTMS IRLSSG PSQI HAM-A HAM-D course history

Patients 1F 516 No 227 2 5 2 F 55 27 No 12 4 5 6 3 M 72 5 Yes Yes 18 11 7 6 4 F 44 10 No 28 13 6 5 5 F 67 4 Yes Yes 24 13 5 8 6F 357 No 134 3 5 7 F 52 34 Yes Yes 24 13 7 9 8 F 63 35 Yes Yes 27 15 5 7 9 F 52 4 No Yes 19 7 5 6 10 M 65 31 Yes Trastal 50 mg 37 17 14 13 11 F 67 41 No Madopar 187.5 mg 29 11 5 8 Pramipexole 0.25 mg 12 F 54 15 Yes Pramipexole 0.25 mg 20 5 8 5 13 F 61 39 Yes Pramipexole 0.5 mg Yes 28 10 8 7 14 F 51 21 No Pramipexole 0.25 mg 28 13 4 7 Clonazepam 1 mg 15 M 59 4 No Estazolam 1 mg Yes 22 12 6 13 Mean + SD 12 F /3M 56.53 ± 9.75 18.87 ± 14.29 23.40 ± 6.52 10.33 ± 4.05 6.00 ± 2.78 7.33 ± 2.61 Controls 1M 62 2F 54 3M 71 4F 50 5F 65 6F 35 7M 71 8M 63 9M 65 10 F 69 11 F 53 12 F 61 13 F 52 14 M 61 Mean ± SD 8 F/6 M 59.43 ± 9.83 p Value 0.18a 0.43b

F, female; IRLSSG, International Restless Leg Study Group Severity Scale; HAM-D, Hamilton Depression Rating Scale; HAM-A, Hamilton Anxiety Rating Scale; M, male; PSQI, Pittsburgh Sleep Quality Index; RLS, restless legs syndrome; rTMS, repetitive transcranial magnetic stimulation; SD, standard deviation. a p Value obtained by two-tailed Pearson χ2 test, which was used for gender comparison between the idiopathic RLS patients and controls. b p Value obtained by a two-sample, two-tailed t-test, which was used for age comparison between the idiopathic RLS patients and controls. C. Liu et al./Sleep Medicine 16 (2015) 785–791 787

B12, folic acid, thyroid hormone, and HbA1c) were normal, and no space and re-sampled to 3-mm isotropic voxels using the normal- patients were on psychotropic medications. ization parameters estimated during unified segmentation. The We also recruited 14 age- and gender-matched right-handed resultant normalized functional images were spatially smoothed with healthy controls (eight females; age range: 35–71 years; mean age: a 4-mm full-width-at-half-maximum (FWHM) Gaussian kernel, and 59.43 ± 9.83 years) (Table 1). The controls had no history of neu- linear trends were removed. Finally, all images were temporally fil- rological problems such as cognitive disorder, psychiatric illness, or tered (0.01–0.08 Hz) to reduce the effects of low-frequency drift and a family history of RLS. Controls were recruited from the commu- high-frequency physiological noise. nity and were evaluated with cognitive and psychiatric scales. All controls had a Mini-Mental State Examination (MMSE) score > 28, 2.4. Functional ALFF analyses a Montreal Cognitive Assessment (MoCA) score ≥ 27, and a Neuro- psychiatric Inventory (NPI) score of 0. This study was approved by We used the Resting-State fMRI Data Analysis Toolkit (REST, the Medical Research Ethics Committee at Xuan Wu Hospital of http://rest.restfmri.net) to calculate the ALFF [7,16]. Briefly, for a given Capital Medical University. Written informed consent was ob- voxel, the time series was first converted to the frequency domain tained from all subjects. using a fast Fourier transform. The square root of the power spectrum was computed and then averaged across 0.01–0.08 Hz. This averaged square root was termed the ALFF. It was then divided by 2.2. Data acquisition the global mean ALFF value for each subject to reduce global effects of variability. Imaging was performed during the daytime, starting from 12:00 when patients were asymptomatic and with empty stomachs. If a 2.5. Intervention: rTMS paradigm patient experienced leg discomfort during scanning, he or she was excluded (two patients were excluded for this reason). MRI data ac- Of the 15 idiopathic RLS patients, seven gave their informed quisition was performed on a Siemens Trio 3-Tesla scanner (Siemens, consent to receive rTMS therapy (see Table 1). Focal rTMS was ad- Erlangen, Germany). Foam padding and headphones were used to ministered through a wind-cooled figure-eight coil (9-cm external limit head motion and to reduce scanner noise. Subjects were in- diameter at each wing) connected to a magnetic stimulator, which structed to move as little as possible, to relax their minds, and to gave a 2.0-Tesla pulse at maximal output (Magstim Super-Rapid; keep their eyes closed without falling asleep. Functional images were Magstim Co., Whitland, UK). The stimulation was directed to the collected axially using an echo-planar imaging (EPI) sequence with leg area of primary motor cortex (M1) with the coil centered at the the following settings: repetition time (TR) = 2000 ms, echo time hot spot for the tibialis anterior (TA) muscle [17,18] (typically from (TE) = 40 ms, flip angle (FA) = 90°, field of view (FOV) = 24 cm, res- 0 to 2 cm lateral to the vertex and from −1 to 2 cm posterior to the olution = 64 × 64 matrix, slices = 28, thickness = 4 mm, voxel vertex). The stimulus intensity was fixed at 120% of active motor size = 3.75 × 3.75 × 4 mm, gap = 1 mm, and band-width = 2232 Hz. threshold (AMT) for the contralateral TA muscle. The AMT was The scan lasted for 478 s and thus included 239 functional volumes defined as the minimum stimulation intensity that produced at least for each subject. A post-scan questionnaire showed that no sub- five motor-evoked potentials with amplitudes ≥ 100 μV over the jects fell asleep during the scan. Three-dimensional T1-weighted course of 10 consecutive trials during voluntary contraction. For this magnetization-prepared rapid gradient echo (MPRAGE) sagittal measurement, the coil was centered over the leg motor area with images were collected using the following parameters: TR = 1900 the handle pointing laterally to induce a lateral-to-medial current ms, TE = 2.2 ms, inversion time (TI) = 900 ms, FA = 9°, resolu- flow in the cortex. Patients wore earplugs during the treatment and tion = 256 × 256 matrix, slices = 176, thickness =1.0 mm, and voxel were seated in a comfortable chair in a reclined position. A head size = 1 × 1 × 1 mm. For patients who received rTMS, a second MRI restraint was used to prevent movement. A coil holder kept the coil scan with the same parameters was performed after completing 14 in a fixed position, and the coil was applied parallel to the vertex rTMS sessions. The time interval between the last rTMS and the and tangentially to a participant’s head surface. A daily session con- second MR scan was more than 24 hours. In principle, the RLS pa- sisted of 20 rTMS trains, half delivered to the left leg area of M1 tients received rTMS treatment in the morning from 08:00 to 10:00 and half to the right. A single train consisted of 50 stimulations de- on 14 consecutive days. On the 15th day, they received the second livered at 5 Hz, and the intertrain interval was 50 seconds. All patients MR scan starting from 12:00, which was the same time of day as were treated daily for 14 consecutive days. A well-trained and qual- the first MR scan. ified physical therapist delivered the rTMS to all patients. Patients who had already been taking dopamine agonists or benzodiaz- 2.3. Data preprocessing epines for at least one month continued to take their medication at the same dosage throughout the 2-week treatment. The IRLSSG, Image preprocessing was carried out using Statistical Paramet- PSQI, HAM-A, and HAM-D scores were assessed before the first ric Mapping (SPM8, http://www.fil.ion.ucl.ac.uk/spm) and Data session and after the 14th session (Table 2). Any adverse effects Processing Assistant for Resting-State fMRI (DPARSF) [13]. The first related to the rTMS procedure were documented for each patient. 10 volumes were discarded to allow for scanner stabilization and participant adaptation to scanning. The remaining scans were first 2.6. Statistical analyses corrected for within-scan acquisition time differences between slices and further realigned to the first volume to correct for head motions. 2.6.1. Between-group differences No participant was excluded for excessive head movements (more To examine between-group differences in ALFF, voxelwise general than 3 mm of translation or 3 degrees of rotation in any direc- linear model (GLM) analysis was performed with age and gender tion). Subsequently, each individual structural image (T1-weighted as covariates. Statistical significance threshold was set at p < 0.05 MPRAGE images) was co-registered to the mean functional image (i.e., height threshold) and cluster size > 2214 mm3 (i.e., extent after motion correction using a linear transformation [14]. The trans- threshold), which corresponded to a corrected p < 0.05. Correc- formed structural images were then segmented into gray matter tion for multiple comparisons was confined to a GM mask (size: (GM), white matter, and cerebrospinal fluid using a unified seg- 1,826,064 mm3) that was generated by thresholding (a threshold mentation algorithm [15]. The motion-corrected functional volumes of 0.2) an a priori gray matter probability map in SPM8 and per- were spatially normalized to Montreal Neurological Institute (MNI) formed by Monte Carlo simulations [19] using the AFNI Alpha 788 C. Liu et al./Sleep Medicine 16 (2015) 785–791

Table 2 Assessment of IRLSSG, PSQI, HAM-A, and HAM-D scores in idiopathic RLS patients (baseline and after 14 sessions of rTMS).

Patient Sex Age (y) IRLSSG PSQI HAM-A HAM-D

Pre Post Pre Post Pre Post Pre Post

1 F 61 28 15 10 6 8 6 7 5 2 M 72 18 9 11 11 7 4 6 5 3 F 67 24 18 13 11 5 5 8 6 4 M 59 22 14 12 8 6 5 13 7 5 F 52 24 12 13 8 7 5 9 7 6 F 63 27 19 15 12 5 5 7 6 7 F 52 19 12 7 6 5 5 6 6 Mean ± SD 5 F/2 M 60.86 ± 7.38 23.14 ± 3.76 14.14 ± 3.53 11.57 ± 2.57 8.86 ± 2.48 6.14 ± 1.22 5 ± 0.58 8 ± 2.45 6 ± 0.82 p < 0.0001a p = 0.0072a p = 0.0472a p = 0.0327a

F, female; IRLSSG, International Restless Legs Syndrome Study Group Severity Scale; HAM-A, Hamilton Anxiety Scale; HAM-D, Hamilton Depression Scale; M, male; PSQI, Pittsburgh Sleep Quality Index; RLS, restless legs syndrome; rTMS, repetitive transcranial magnetic stimulation; SD, standard deviation. a p Value obtained by paired t-test.

Sim program (http://afni.nimh.nih.gov/pub/dist/doc/manual/ 3.2. Comparisons of ALFF between the drug-naive RLS patient and AlphaSim.pdf). control groups

Fig. 1B shows a similar pattern when comparing ALFF values 2.6.2. Within-group differences for the effect of rTMS between the nine drug-naive RLS patient group and the healthy A paired t-test was performed to examine changes in the ALFF control group. Lower ALFF values were again found in the paracen- before and after rTMS treatment. Statistical signiﬁcance threshold tral lobule, precuneus, supplementary motor area, and occipital lobe. was set to p < 0.05 and cluster size > 2214 mm3, which corre- The lower ALFF values in the right precentral gyrus and higher ALFF sponded to a corrected p < 0.05. Correction for multiple comparisons values in the insula, hippocampus, and left posterior parietal area was conﬁned to the GM mask. also survived the height threshold but not the extent threshold (1026 mm3). 3. Results

3.3. Comparisons of ALFF between RLS patients before rTMS 3.1. Comparisons of the ALFF between RLS patient and control treatment and healthy controls groups

Fig. 2A shows the differences in ALFF values between the healthy Fig. 1A shows the differences in the ALFF between the group of controls and the seven RLS patients who subsequently received rTMS 15 RLS patients and the group of healthy controls. RLS patients therapy. We found lower ALFF values in the occipital lobe and higher showed signiﬁcantly lower ALFF values in the sensorimotor system, ALFF values in the hippocampus, rectus, left insula, superior frontal including the paracentral lobule, precuneus, superior parietal gyrus, gyrus, and left posterior parietal area. Lower ALFF in the SMA and supplementary motor area (SMA), right precentral gyrus, right post- right precentral gyrus also survived the height threshold but not central gyrus, and visual processing system, including middle the extent threshold (837 mm3). occipital gyrus, calcarine sulcus, cuneus, fusiform gyrus, and right inferior temporal gyrus. We also found that these patients had sig- niﬁcantly higher ALFF values in the insula, parahippocampal and 3.4. Changes in ALFF values after rTMS treatment hippocampal gyri, inferior frontal gyrus, rectus, left inferior parietal gyrus, left superior parietal gyrus, left angular gyrus, and Fig. 2B shows regions in which ALFF values changed after rTMS brainstem. treatment in the seven RLS patients. Increased ALFF values were

Fig. 1. Difference in amplitude of low-frequency ﬂuctuations (ALFF) between the idiopathic restless legs syndrome (RLS) patients and the healthy control (HC) group. (A) Z-statistic difference map between the 15 RLS patients and the HC group. Statistical signiﬁcance threshold was set at p < 0.05 (ie, height threshold) and cluster size >2214 mm3 (ie, extent threshold), which corresponded to a corrected p value of <0.05. (B) Z-statistic difference map between the nine drug-naive RLS patients and the HC group. Lower ALFF values were again found in the paracentral lobule, precuneus, supplementary motor area, and occipital lobe. Lower ALFF in the right precentral gyrus and higher ALFF in the insula, hippocampus, and left posterior parietal area survived the height threshold (p < 0.05) but not the extent threshold (1026 mm3). C. Liu et al./Sleep Medicine 16 (2015) 785–791 789

Fig. 2. (A) Difference in amplitude of low-frequency ﬂuctuations (ALFF) between the seven restless legs syndrome (RLS) patients before repetitive transcranial magnetic stimulation (rTMS) treatment and the healthy control (HC) group. We found lower ALFF values in the occipital lobe and higher ALFF values in the hippocampus, rectus, left insula, superior frontal gyrus, and left posterior parietal area. Decreased ALFF in the SMA and right precentral gyrus survived the height threshold (p < 0.05) but not the extent threshold (837 mm3). (B) Changes in ALFF after TMS treatment. Increased ALFF values were primarily in the sensorimotor cortex and the occipital lobe, and decreased values were primarily in the superior frontal gyrus, middle frontal gyrus, and right angular gyrus. (C) Difference in ALFF between the seven RLS patients after rTMS treatment and the HC group. Regions with higher ALFF in RLS patients included the rectus, left parahippocampal gyrus, hippocampal gyrus, left inferior parietal gyrus, and superior parietal gyrus. ALFF values in the sensorimotor cortex and occipital lobe did not differ between groups after rTMS treatment. Higher ALFF in the left insula and lower ALFF in right fusiform gyrus and right cuneus survived the height threshold but not the extent threshold (567 mm3).

primarily in the sensorimotor cortex and the occipital lobe, and de- 3.6. Changes in IRLSSG score after rTMS treatment creased values were primarily in the superior frontal gyrus, middle frontal gyrus, and right angular gyrus. Fig. 3 shows IRLSSG Rating Scale scores before and after delivering rTMS to the leg area of M1 daily for 14 days. Data analysis

3.5. Comparison of ALFF values between RLS patients after rTMS treatment and healthy controls

Fig. 2C shows the differences in ALFF values between the group of seven RLS patients after rTMS treatment and the group of healthy controls. Regions with higher ALFF values in RLS patients compared with healthy controls included the rectus, left parahippocampal gyrus, hippocampal gyrus, left inferior parietal gyrus, and superi- or parietal gyrus. ALFF values in the sensorimotor cortex and occipital lobe did not differ between the RLS patient and control groups after rTMS treatment. Higher ALFF in left insula and lower ALFF in right fusiform gyrus and right cuneus survived the height threshold but Fig. 3. International Restless Legs Syndrome Study Group (IRLSSG) Rating Scale scores not the extent threshold (567 mm3). before and after repetitive transcranial magnetic stimulation (rTMS). *p < 0.0001. 790 C. Liu et al./Sleep Medicine 16 (2015) 785–791 revealed that scores significantly decreased after treatment (before: sensorimotor regions and the occipital lobe and hyperactivity in the 23.14 ± 3.76; after: 14.14 ± 3.53; p < 0.0001). insula and posterior parietal areas both contribute to RLS pathogenesis. 4. Discussion In our study, patient compliance was high; no adverse effects were observed; and high-frequency rTMS stimulation over leg motor- By measuring the ALFF, we found brain regions in which activ- cortex relieved RLS symptoms. However, the mechanism underlying ity differed between the idiopathic RLS patient and the normal the therapeutic effects is still unclear. This could be explained by control groups, even when the patients were experiencing asymp- changes in the dopaminergic system. High-frequency rTMS over M1 tomatic periods. The major findings included the following: was reported to induce a focal release of endogenous dopamine (1) Patients showed lower ALFF in the sensorimotor areas and the within the ipsilateral dorsal striatum (putamen, caudate nucleus), occipital lobe, and higher ALFF values in the insula, parahippocampal probably by activating cortico-striatal projections [32]. In addi- and hippocampal gyri, left posterior parietal areas, and brain- tion, dopamine acting on D2Rs plays a role in descending inhibitory stem; and (2) after delivering high-frequency rTMS to M1 leg cortex, control at several central sites, including dorsal striatum, hypo- symptoms improved as assessed by the IRLSSG Rating Scale, and thalamus A11-cell group, and spinal cord [33]. The increased release the ALFF values in the sensorimotor regions and occipital lobe sig- of endogenous dopamine could act on these sites to enhance de- nificantly increased. scending inhibition and to prevent abnormal somatosensory Clinical and pharmacological observations suggest that dopa- processing. Furthermore, the antinociceptive effect induced by stim- minergic hypoactivity may play a role in the pathophysiology of RLS ulation of the OC likely also results from the activation of a [20]. In our study, we found hypoactivation in M1 and SMA in id- descending inhibitory pathway [34]. In this way, normalizing M1 iopathic RLS patients. We speculate that this results from and occipital cortical activity could be a way to relieve unpleasant dopaminergic hypoactivity and cortical deafferentation of the basal sensation in the legs, whereas abnormal function in these brain ganglia–thalamo-cortical circuit [21]. The low activity in senso- regions likely has a pathophysiological significance in idiopathic RLS rimotor and occipital cortices during asymptomatic periods may patients. contribute to abnormal somatosensory processing, leading to dis- The abnormally high ALFF values in the rectus, left para- comfort in the leg. Many studies regarding pain have found that hippocampal and hippocampal gyri, and posterior parietal cortex exciting the primary motor cortex can have an antinociceptive role did not change significantly after rTMS. This could have been because through the activation of a descending inhibitory pathway [22–25]. the stimulation did not penetrate to deeper brain regions such as Because motor responses are closely linked to visual stimuli, visual the rectus and the parahippocampal and hippocampal gyri. Another information processing in the occipital lobe is an important part of reason could be that the duration of stimulation was not long enough the sensorimotor network. A positron emission tomography study for normalization of functional activities in these regions to occur. conducted in resting fibromyalgic patients revealed regional cere- Further studies that increase the number of rTMS sessions should bral blood flow that was lower in the occipital cortex (OC) than in be conducted to further elevate activity in the motor cortex, and that of controls [26]. Injection of glutamate into the OC also reduced thus increase descending inhibition and help prevent abnormal so- pain in the rat tail-flick test [27]. Because glutamate excites neu- matosensory processing. ronal cell bodies but not fibers, this suggests that the increased Several limitations of our study should be addressed. First, only neuronal activity in the OC results in antinociception [26]. Thus, we seven patients received stimulation, and no sham stimulation control speculate the low activity in motor and occipital cortices of idio- was included. Although the improvement of IRLSSG Rating Scale pathic RLS patients may weaken the inhibitory effects of descending scores in RLS patients was between 25% and 50%, we cannot exclude pathways, leading to the abnormal central somatosensory processing. the possibility that this was a placebo effect. Moreover, several pa- Regions with higher ALFF values were found in the insula, tients were medicated while receiving rTMS treatment, and thus parahippocampal gyrus, hippocampal gyrus, and left posterior pa- a potential cumulative effect of magnetic stimulation and medica- rietal areas. In RLS patients, changes have been found in the binding tion cannot be excluded. Second, most patients had abnormal PSQI potential of dopamine D2-receptors (D2Rs) in the insula [19]. In- scores, implying sleep deprivation [35,36]. Because PSQI scores were creased activation in the left pars opercularis was observed through not obtained from controls, we cannot rule out the possibility that fMRI during nighttime episodes of sensory leg discomfort and pe- the lower ALFF in the visual processing system was in part related riodic limb movements [5]. Functionally, the insula has been to sleep deprivation in RLS patients. Third, the current dataset is postulated to play a key role in maintaining homeostasis by moni- cross-sectional and does not allow us to examine ALFF-related toring and integrating interoceptive visceral and somatic feelings, dynamic changes with RLS circadian variations; future follow-up and translating these to conscious emotional perceptions [28,29]. studies are warranted to examine RLS circadian variations. Finally, Thus, it integrates sensory and visceral signals from peripheral re- the sample size was small, and further validation using a large sample ceptors. The high activity of insular and posterior parietal areas may is therefore necessary. This project is still ongoing, and we are plan- be related to abnormal central somatosensory processing. To date, ning to make these adjustments and to observe the cumulative effect most investigations of aversive conditioning have highlighted spe- of increasing the number of repetitive sessions. cific contributions from the amygdala and parahippocampal gyrus In conclusion, abnormally low spontaneous activity in the sen- [30]. Findings show that patients with parahippocampal gyrus lesions sorimotor cortex and the occipital lobe may be involved in the do not habituate to mildly aversive somatosensory stimuli and are pathogenesis of idiopathic RLS. High-frequency (5-Hz) rTMS di- impaired in their ability to modify neural responses to them [31]. rected toward M1 (leg area) can mitigate this problem and relieve Thus, we speculate that in RLS patients, higher ALFF in the some symptoms of idiopathic RLS. rTMS may promote the release parahippocampal and hippocampal gyri may help habituation to un- of dopamine and, in turn, enhance the descending inhibitory pleasant and aversive stimuli coming from the legs. pathway and prevent abnormal central somatosensory processing. Results after comparing ALFF between drug-naive patients and controls followed a similar pattern. Although significant changes Conflict of interest were not seen in exactly the same regions, this was likely due to the small sample size (nine RLS patients). Indeed, after reducing the The ICMJE Uniform Disclosure Form for Potential Conflicts of In- significance threshold, we did find the same pattern of changes in terest associated with this article can be viewed by clicking on the the same brain regions. It appears that hypoactivity in the following link: http://dx.doi.org/10.1016/j.sleep.2014.12.029. C. Liu et al./Sleep Medicine 16 (2015) 785–791 791

Acknowledgement [17] Terao Y, Furubayashi T, Okabe S, et al. Interhemispheric transmission of visuomotor information for motor implementation. Cereb Cortex 2005;15:1025–36. This work was supported by the National Natural Science Foun- [18] Terao Y, Furubayashi T, Okabe S, et al. Modifying the cortical processing for dation of China, Grant Nos. 81271494, 81301138, 81030028, and motor preparation by repetitive transcranial magnetic stimulation. J Cogn 81225012. Neurosci 2007;19:1556–73. [19] Ledberg A, Akerman S, Roland PE. Estimation of the probabilities of 3D clusters in functional brain images. Neuroimage 1998;8:113–28. References [20] Simon C, Sven EP, Robert A, et al. Support for dopaminergic hypoactivity in restless legs syndrome: a PET study on D2-receptor binding. Brain 2006;129:2017–28. [1] Allen RP, Picchietti D, Hening WA, et al. Restless legs syndrome: diagnostic [21] Buhmann C, Glauche V, Stuèrenburg HJ, et al. Pharmacologically modulated criteria, special considerations, and epidemiology: a report from the restless fMRI-cortical responsiveness to levodopa in drug-naive hemiparkinsonian legs syndrome diagnosis and epidemiology workshop at the National Institutes patients. Brain 2003;126:451–61. of Health. Sleep Med 2003;4:101–19. [22] Reidler JS, Mendonca ME, Santana MB, et al. Effects of motor cortex modulation [2] Trenkwalder C, Paulus W, Walters AS. The restless legs syndrome. Lancet Neurol and descending inhibitory systems on pain thresholds in healthy subjects. J Pain 2005;4:465–75. 2012;13:450–8. [3] Bucher SF, Seelos KC, Oertel WH, et al. Cerebral generators involved in the [23] Fregni F, Boggio PS, Lima MC, et al. A sham-controlled, phase II trial of pathogenesis of the restless legs syndrome. Ann Neurol 1997;41:639–45. transcranial direct current stimulation for the treatment of central pain in [4] Margariti PN, Astrakas LGT, Souli SG, et al. Investigation of unmedicated early traumatic spinal cord injury. Pain 2006;122:197–209. onset restless legs syndrome by voxel-based morphometry, T2 relaxometry, and [24] Mendonca ME, Santana MB, Baptista AF, et al. Transcranial DC stimulation in functional MR imaging during the night-time hours. AJNR Am J Neuroradiol fibromyalgia: optimized cortical target supported by high resolution 2012;33:667–72. computational models. J Pain 2011;12:610–17. [5] Spiegelhalder K, Feige B, Paul D, et al. Cerebral correlates of muscle tone [25] Fenton BW, Palmieri PA, Boggio P, et al. A preliminary study of transcranial direct fluctuations in restless legs syndrome: a pilot study with combined functional current stimulation for the treatment of refractory chronic pelvic pain. Brain magnetic resonance imaging and anterior tibial muscle electromyography. Sleep Stimul 2009;2:103–7. Med 2008;9:177–83. [26] Wik G, Fischer H, Bragée B, et al. Retrosplenial cortical activation in the [6] Biswal B, Yetkin FZ, Haughton VM, et al. Functional connectivity in the motor fibromyalgia syndrome. Neuroreport 2003;14:619–21. cortex of resting human brain using echo-planar MRI. Magn Reson Med [27] Fries W, Zieglgansberger W. A method to discriminate axonal from cell body 1995;34:537–41. activity and to analyze ‘silent cells’. Exp Brain Res 1974;21:441–5. [7] Zang YF, He Y, Zhu CZ, et al. Altered baseline brain activity in children with [28] Craig AD. The sentient self. Brain Struct Funct 2010;214:563–77. ADHD revealed by resting-state functional MRI. Brain Dev 2007;29:83–91. [29] Craig AD. Significance of the insula for the evolution of human awareness of [8] Lui S, Huang X, Chen L, et al. High-field MRI reveals an acute impact on brain feelings from the body. Ann NY Acad Sci 2011;1225:72–82. function in survivors of the magnitude 8.0 earthquake in China. Proc Natl Acad [30] Andreea OD, Mahesh M, Jimmy J, et al. Dopamine-induced changes in neural SciUSA2009;106:15412–17. network patterns supporting aversive conditioning. Brain Res 2010;1313:143– [9] Chen R, Classen J, Gerloff C, et al. Depression of motor complex excitability by 61. low-frequency transcranial magnetic stimulation. Neurology 1997;48:1398–403. [31] Ousdal OT, Jensen J, Server A, et al. The human amygdala is involved in general [10] Boroojerdi B, Prager A, Muellbacher W, et al. Reduction of human visual cortex behavioral relevance detection: evidence from an event-related functional excitability using 1 Hz transcranial magnetic stimulation. Neurology magnetic resonance imaging Go–NoGo task. Neuroscience 2008;156:450–5. 2000;54:1529–31. [32] Strafella AP, Ko JH, Grant J, et al. Corticostriatal functional interactions in [11] Rotenberg A, Muller P, Birnbaum D, et al. Seizure suppression by EEG-guided Parkinson’s disease: a rTMS/[11C]raclopride PET study. Eur J Neurosci repetitive transcranial magnetic stimulation in the rat. Clin Neurophysiol 2005;22:2946–52. 2008;119:2697–702. [33] Wood PB. Role of central dopamine in pain and analgesia. Expert Rev Neurother [12] Fitzgerald PB, Fountain S, Daskalakis ZJ. A comprehensive review of the effects 2008;8:781–97. of rTMS on motor cortical excitability and inhibition. Clin Neurophysiol [34] Zhang JH, Cao XD, Lie J, et al. Neuronal specificity of needling acupoints at same 2006;117:2584–96. meridian: a control functional magnetic resonance imaging study with [13] Yan CG, Zang YF. DPARSF: a MATLAB toolbox for “pipeline” data analysis of electroacupuncture. Acupunct Electrother Res 2007;32:179–93. resting-state fMRI. Front Syst Neurosci 2010;4. [35] Habecka C, Rakitina BC, Moeller J, et al. An event-related fMRI study of the [14] Collignon A, Maes F, Delaere D, et al. Automated multi-modality image neurobehavioral impact of sleep deprivation on performance of a delayed- registration based on information theory. Inf Process Med Imag 1995;263–74. match-to-sample task. Cogn Brain Res 2004;18:306–21. [15] Ashburner J, Friston KJ. Unified segmentation. Neuroimage 2005;26:839–51. [36] Anna SU, Antti H, Sami H, et al. Preliminary findings of proton magnetic [16] Song XW, Dong ZY, Long XY, et al. REST: a toolkit for resting-state functional resonance spectroscopy in occipital cortex during sleep deprivation. Psychiatry magnetic resonance imaging data processing. PLoS ONE 2011;6:e25031. Res Neuroimaging 2006;147:41–6.